---
base_model: ukisai/Swift-Qwen3.8-27b
base_model_relation: quantized
library_name: transformers
license: other
license_name: swift-open-license-1.0
license_link: https://ukisai.com/news/introducing-swift
tags:
  - int4
  - w4a16
  - compressed-tensors
  - auto-round
  - vllm
  - qwen3
---

# Swift-Qwen3.8-27b-W4A16-AutoRound

Quantized [ukisai/Swift-Qwen3.8-27b](https://huggingface.co/ukisai/Swift-Qwen3.8-27b)
with [Intel AutoRound](https://github.com/intel/auto-round) and tested in vLLM. It appears to work!

## Recipe

Copied the work of [dbirks/Qwen3.8-27B-W4A16-AutoRound](https://huggingface.co/dbirks/Qwen3.8-27B-W4A16-AutoRound) except with embedding unquantized.

## License

Base model distributed under [UkisAI's Swift Open License v1.0](https://ukisai.com/news/introducing-swift). This quantization probably inherits those terms, but I'm not a lawyer.